[R-sig-ME] Cholmod warning

[Douglas Bates]

On Sun, Jul 24, 2011 at 8:22 AM, Iker Vaquero Alba <karraspito at yahoo.es> wrote:

>    Dear list:

>    I am implementing a model with lmer in which the dependent variable is the number of chicks fledged by a number of barn swallow couples. I am using the simplification method, and have realized that the order in which different elements are introduced in the model has a great important, to the extent of making a "false convergence" warning message disappear. After doing all the simplification, I arrive to this model:

>    s.hatchcoltailmodel1.76<-lmer(nhatch1~sex+tlength+brithr+briventral+cond+weatherpc1+weatherpc1:briventral+weatherpc1:cond+(1|site/pair)+(1|year),family=poisson,na.action=na.omit)

>    This one, the LAST one (and the simplest) gives me this error message:

>    Mensajes de aviso perdidos
> 1: In mer_finalize(ans) :
>   Cholmod warning 'not positive definite' at file:../Cholesky/t_cholmod_rowfac.c, line 432
> 2: In mer_finalize(ans) :
>   Cholmod warning 'not positive definite' at file:../Cholesky/t_cholmod_rowfac.c, line 432
> 3: In mer_finalize(ans) : false convergence (8)

>    No matter the order in which I write the elements.

>    I really cannot understand how it is possible that I have been able to make all the simplification from the maximal model (18 parameters), and after a long and careful process, exactly when I arrive to my minimal model (8 parameters and only 2 interactions) is when everything collapses. It is really annoying.

Add the verbose=TRUE option to see if you can determine which of the
parameters is going off to unusual values.  It is entirely possible
that your model is overparameterized.

>
>    Any suggestions, comments or ideas'll be much appreciated.
>
>    Thank you very much.
>
>
>   ________________________________________
>
>    Iker Vaquero-Alba
>    Centre for Ecology and Conservation
>    Daphne du Maurier Building
>    University of Exeter, Cornwall Campus
>    Treliever Road
>    TR10 9EZ Penryn
>    U.K.
>
>    http://biosciences.exeter.ac.uk/cec/staff/postgradresearch/ikervaquero-alba/
>        [[alternative HTML version deleted]]
>
>
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>
>